RESUMO
BACKGROUND: We describe the decentralisation of antiretroviral therapy (ART) alongside Option B+ roll-out in public healthcare facilities in the Lablite project in Uganda. Lessons learned will inform programmes now implementing universal test and treat (UTT). METHODS: Routine data were retrospectively extracted from ART registers between October 2012 and March 2015 for all adults and children initiating ART at two primary care facilities (spokes) and their corresponding district hospitals (hubs) in northern and central Uganda. We describe ART initiation over time and retention and use of Cox models to explore risk factors for attrition due to mortality and loss to follow-up. Results from tracing of patients lost to follow-up were used to correct retention estimates. RESULTS: Of 2100 ART initiations, 1125 were in the north, including 944 (84%) at the hub and 181 (16%) at the spokes; children comprised 95 (10%) initiations at the hubs and 14 (8%) at the spokes. Corresponding numbers were 642 (66%) at the hub and 333 (34%) at the spokes in the central region (77 [12%] and 22 [7%], respectively, in children). Children <3 y of age comprised the minority of initiations in children at all sites. Twenty-three percent of adult ART initiations at the north hub were Option B+ compared with 45% at the spokes (25% and 65%, respectively, in the central region). Proportions retained in care in the north hub at 6 and 12 mo were 92% (95% CI 90 to 93) and 89% (895% CI 7 to 91), respectively. Corresponding corrected estimates in the north spokes were 87% (95% CI 78 to 93) and 82% (95% CI 72 to 89), respectively. In the central hub, corrected estimates were 84% (95% CI 80 to 87) and 78% (95% CI 74 to 82), and were 89% (95% CI 77.9 to 95.1) and 83% (95% CI 64.1 to 92.9) at the spokes, respectively. Among adults newly initiating ART, being older was independently associated with a lower risk of attrition (adjusted hazard ratio [aHR] 0.93 per 5 y [95% CI 0.88 to 0.97]). Other independent risk factors included initiating with a tenofovir-based regimen vs zidovudine (aHR 0.60 [95% CI 0.46 to 0.77]), year of ART initiation (2013 aHR 1.55 [95% CI 1.21 to 1.97], ≥2014 aHR 1.41 [95% CI 1.06 to 1.87]) vs 2012, hub vs spoke (aHR 0.35 [95% CI 0.29 to 0.43]) and central vs north (aHR 2.28 [95% CI 1.86 to 2.81]). Independently, patient type was associated with retention. CONCLUSIONS: After ART decentralisation, people living with human immunodeficiency virus (HIV) were willing to initiate ART in rural primary care facilities. Retention on ART was variable across facilities and attrition was higher among some groups, including younger adults and women initiating ART during pregnancy/breastfeeding. Interventions to support these groups are required to optimise benefits of expanded access to HIV services under UTT.